Ratiometric Fluorescent Sensors Illuminate Cellular Magnesium Imbalance in a Model of Acetaminophen-Induced Liver Injury

Magnesium(II) plays catalytic, structural, regulatory, and signaling roles in living organisms. Abnormal levels of this metal have been associated with numerous pathologies, including cardiovascular disease, diabetes, metabolic syndrome, immunodeficiency, cancer, and, most recently, liver pathologies affecting humans. The role of Mg2+ in the pathophysiology of liver disease, however, has been occluded by concomitant changes in concentration of interfering divalent cations, such as Ca2+, which complicates the interpretation of experiments conducted with existing molecular Mg2+ indicators. Herein, we introduce a new quinoline-based fluorescent sensor, MagZet1, that displays a shift in its excitation and emission wavelengths, affording ratiometric detection of cellular Mg2+ by both fluorescence microscopy and flow cytometry. The new sensor binds the target metal with a submillimolar dissociation constant—well suited for detection of changes in free Mg2+ in cells—and displays a 10-fold selectivity against Ca2+. Furthermore, the fluorescence ratio is insensitive to changes in pH in the physiological range, providing an overall superior performance over existing indicators. We provide insights into the metal selectivity profile of the new sensor based on computational modeling, and we apply it to shed light on a decrease in cytosolic free Mg2+ and altered expression of metal transporters in cellular models of drug-induced liver injury caused by acetaminophen overdose.

All reagents were purchased from commercial sources and used as received. Solvents were purified and degassed by standard procedures. Bromomethyl acetate was prepared as described by Grynkiewicz  Coupling constants are reported in Hz. High resolution mass spectra (HRMS) were acquired on an Agilent 6224 Accurate-Mass TOF LC/MS using APCI or ESI ionization. Reversed phase HPLC analyses were conducted using an Agilent 1260 Infinity system with UV-vis and fluorescence detection using a Zorbax Extend C18 reversed phase column (4.6×50 mm, 1.8 µm particle size) and a gradient from 10% to 100% acetonitrile/water (+ 0.1% trifluoroacetic acid). Low resolution mass spectra (LRMS) were acquired on an Agilent 6120 Single Quadrupole LCMS spectrometer using ESI ionization and a Kinetex C18 reversed phase column (3×50 mm, 2.6 µm particle size) and a gradient of 5% to 100% acetonitrile/water (+ 0.1% formic acid).
After 2 h, the reaction mixture was transferred quantitatively to a volumetric flask and diluted to S14 2 mL with PIPES buffer (pH 7.04) to yield a 14.33 mM stock solution of MagDMA for spectroscopic studies. Reaction completion was verified by HPLC.

Synthesis of methyl 2-amino-5-bromobenzoate (8)
The brominated anthranilate was prepared following a modified procedure from Chaudhuri et. al. 4 Briefly, to a solution of methyl anthranilate (1 mL, 7 mmol, 1 equiv.) in diethyl ether (3 M) was added dioxane dibromide (3 g, 1.15 equiv.) at 0 °C. The reaction was then stirred and allowed to gradually warm to room temperature over 50 min, over which a precipitate formed. The waxy brown solid was collected by vacuum filtration, washed with cold ether (3×) and DCM (1×), and dried under vacuum to give the desired product in a quantitative yield. Characterization matched reported values. 4
Azetidine (200 μL, 1.4 equiv.) was added followed by copper iodide (50 mol% with respect to the bromobenzoate). The reaction was heated to 90 °C and stirred overnight. After 18 h, the reaction was diluted with ethyl acetate and washed with water (2×) and brine. The aqueous layers were combined and extracted with ethyl acetate (2×). The organic layers were combined and dried over S16 Na2SO4. Evaporation under reduced pressure produced a brown solid that was purified by flash chromatography (SiO2, 20-30% ethyl acetate/hexanes, Rf = 0.17 in 20% ethyl acetate/hexanes) to yield the desired product as a bright yellow solid (204.7 mg, 45%).
The reaction mixture was concentrated and purified by column (SiO2, 20% ethyl acetate/hexanes, Rf = 0.23) to give the desired product as an orange solid (195.0 mg, 55%).
Evaporation under reduced pressure produced a brown solid that was purified by column (SiO2, 30% ethyl acetate/hexanes Rf = 0.17) to yield the title product as a yellow solid (32.3 mg, 31%).

General spectroscopic details
All aqueous solutions were prepared using deionized water having a resistivity of ≥ 18.  To determine the affinity of the sensor for Zn 2+ , titrations were conducted using an EGTA buffer system consisting of 10 mM EGTA and the total ZnCl2 concentration varying from 0 to 1 mM. The concentration of free Zn 2+ was calculated from the total metal concentration by solving eq S2 with an apparent binding constant for Zn 2+ -EGTA of logK ' ZnEGTA = 8.24 at pH 7.0.

Determination of metal dissociation constants
The apparent binding constant was calculated from the absolute stability constant logKZnEGTA = 12.6 (25 °C, μ = 0.1) listed in Martell and Smith, 7 using Schwarzenbach's method, according to the following equation: To determine the apparent affinity of MagZet1 for ATP-bound Mg 2+ , plots of the fluorescence ratio against [Mg 2+ ]T were fitted assuming binary and ternary complex formation, in addition to competition between MagZet1 and ATP for free Mg 2+ using the following equation: Where R represents the ratio of fluorescence intensity at the metal bound (λ1 = 530 nm) and metal free (λ2 = 500 nm) wavelengths excited at 390 nm (R=Iλ1/Iλ2); Sf1, Sb1, and ST1 are proportionality coefficients for the fluorescence at λ1 of the free sensor, bound sensor, and ternary complex, respectively; and Sf2, Sb2, and ST2 are proportionality coefficients for the fluorescence at λ2 of the

Computational details
DFT calculations on all molecules were carried out using the Gaussian 16 program package 8 applying both a tightened self-consistent field convergence criterion (10 -9 -10 -10 au) and an improved optimization threshold (10 -5 au on average forces). Ground state geometry optimized structures of MagZet1 complexed with Ca 2+ and Mg 2+ were obtained and the corresponding vibrational spectra were computed and analyzed to verify the conversion to a minimum by the absence of imaginary frequencies. All calculations used the same DFT integration grid, the ultrafine pruned (99590) grid, as well as the M06-2X hybrid exchange-correlation functional, 9 which has been shown to be a good choice for investigating structures and excited states push-pull type dyes. [10][11][12][13] Structural parameters were obtained with the extended 6-311+G(d,p) atomic basis set. Bulk solvation effects of the aqueous medium were estimated using the polarizable continuum model (PCM) 14 that was systematically applied to all computational steps. In all calculations, explicit water molecules were added to complete the metal coordination sphere (Mg, pseudo-S27 octahedral; Ca, 8-coordinate) as is common for these metals and observed in crystal structures.
Optimized structures were visualized with and graphics were generated using the Mercury

Cell culture
HeLa cells were grown in a culture system containing Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) at 37 °C in a 5% CO2 atmosphere according to ATCC guidelines. Cells were subcultured or the medium was changed twice a week.
THLE-2 cells were cultured in a system containing Bronchial Epithelial Cell Basal Medium (BMEM) supplemented with growth factor BulletKit (Lonza) from which gentamycin/amphotericin (GA) and epinephrine are discarded. The medium is then further supplemented with epidermal growth factor (EGF, 5 ng/mL), phosphoethanolamine (70 ng/mL), 10% FBS, 1% glutamine, and 1% Penicillin and 1% Streptomycin. THLE-2 cells were grown on plates precoated with a mixture of 0.01 mg/mL fibronectin, 0.01 mg/mL bovine serum albumin, and 0.03 mg/mL collagen. Cells were sub-cultured or the medium was changed twice a week.

General imaging details
Confocal images were obtained using a 63× oil immersion objective (NA = 1.4) in a Zeiss 880 inverted laser scanning confocal microscope equipped with a Zeiss AxioObserver inverted stand, environmental chamber for temperature and stage insert for CO2 control, and Definite Focus autofocusing system; or in a Leica TCS SP8 confocal laser scanning microscope with GaAsP hybrid detectors and OkoLab stage-top environmental chamber. A 405 nm excitation laser line was used and emission was collected from 470 to 520 for the free form and 520 to 582 for the metal bound sensor. Image processing and ratio measurements were conducted using ImageJ. 16 For fluorescence ratio images, background correction and thresholding was applied to individual images prior to ratio calculation.

Imaging free Mg 2+ in HeLa cells with MagZet1AM
HeLa cells, grown as described above, were plated on 3.5 cm glass bottom imaging dishes (MatTek). Cells were grown to ~80% confluency prior to imaging. Cells were washed with warm DMEM without FBS (1×2 mL) and then bathed in DMEM without FBS, containing 5 µM MagZet1AM added from a stock solution of the sensor suspended in 20% Pluronic F127 according to manufacturer protocol (final concentration of Pluronic in the plate was 0.01%). Sensor loading was conducted for 30 min at room temperature, protected from light. After this period, the medium was replaced with fresh DMEM without FBS (2 mL) and the cells were left at room temperature protected from light for an additional 30 min to allow for complete de-esterification of the sensor.
The cells were then washed with HBSS without divalent cations (1×2 mL) and were then bathed in HBSS without cations (1 mL). To confirm the sensors response to Mg 2+ , cells on the microscope stage were supplemented with MgCl2 and the nonfluorescent ionophore 4-Br-A23187 (Molecular Probes) for a total concentration of 50 mM and 10 µM, respectively. Additionally, cells on the microscope stage were supplemented with an aqueous solution of EDTA free acid and the nonfluorescent ionophore 4-Br-A23187 (Molecular Probes) for a total concentration of 50 mM and 10 µM, respectively, to deplete Mg 2+ levels. Images were collected immediately before treatment with Mg 2+ (or EDTA) and every 5 min afterward for 1 h. For comparison, images were acquired and processed under identical conditions.

Control experiments to rule out detection of Ca 2+ or Zn 2+
To test for possible interference from other cations in cellulo, HeLa cells, loaded with MagZet1AM as described above and bathed with HBSS without metal cations, were treated on the microscope stage with Ca 2+ -selective chelator BAPTA-AM for a final concentration of 10 μM. Images were collected every 5 min for 30 min. To rule out interference from Zn 2+ , cells on the microscope stage were treated with Zn 2+ -selective chelator TPA for a final concentration of 10 μM and imaged 5 min after treatment. For comparison, images were acquired and processed under identical conditions.

Imaging free Mg 2+ in paracetamol-treated THLE-2 cells with MagZet1AM
THLE-2 cells, cultured as described above, were plated on 3.5 cm glass bottom imaging dishes (MatTeK) and were grown to ~80% confluency prior to treatment and imaging. Paracetamol

Flow cytometry
For flow cytometry experiments, MagZet1 was detected using a BD LSRII flow cytometer with a 407 nm violet excitation laser. Emission from the "free" sensor was collected using a BV421 filter set with a 450/50 bandpass filter and the signal from the "bound" sensor was collected using a BV510 filter set with a 505 nm LP dichroic mirror and a 530/30 nm bandpass filter. Prior to running samples for analysis, a "Mg 2+ saturated" sample, containing 50 mM MgCl2 and 10 μM the nonfluorescent ionophore 4-Br-A23187 in the medium, and a "Mg 2+ depleted" sample, containing 100 mM EDTA acid and 10 μM the nonfluorescent ionophore 4-Br-A23187 in the medium, were used to determine the full dynamic range of the sensor. The fluidics were then purged for several minutes with 70% ethanol and sheath to remove traces of ionophore.

S32
HeLa cells or THLE-2 cells were grown on 10 cm dishes, trypsinized, and collected by centrifugation. Cell pellets were then resuspended in DMEM without FBS or phenol red to a final concentration of ~1×10 6 cells/mL. Cells were stained in suspension using medium containing 5 µM MagZet1AM added from a stock solution of the sensor suspended in 20% Pluronic F127 according to manufacturer protocol (final concentration of Pluronic in solution was 0.01%) for 30 min at room temperature protected from light. After this period, cells were collected by centrifugation, resuspended in DMEM without FBS or phenol red, and left at room temperature protected from light for an additional 30 min to allow for de-esterification of the sensor. Cells were then pelleted and resuspended in DMEM without phenol red and FBS. Samples were aliquoted and kept on ice. Cell suspensions were gently vortexed and warmed to room temperature immediately prior to data acquisition. FCS files were exported and analyzed in FlowJo v10.9.
Bivariate scatterplots were analyzed and gated according to the following scheme: time gate for cleaning > FSC v. SSC to gate out debris > FSC-H v FSC-A to gate out doublets > SSC v. BV421 or BV510 for mean fluorescence intensities (MFIs) for the metal-bound and metal-free forms of the sensor, respectively. Fluorescence minus one (FMO) samples were used to set gates for "stained" cells.

Statistical analysis
Changes in fluorescence ratio upon treatment were analyzed using the population comparison feature in FlowJo v10.9, which is based on a Cox Chi Squared approach 17 but using probability binning. Histograms of fluorescence ratio (510nm/421nm) of each sample was compared with a control, vehicle-treated sample analyzed and processed using the same protocol and parameters.
Chi-squared values from pair-wise population comparisons were used to calculate p-values reported.

RNA isolation and first strand cDNA generation
THLE-2 cells were grown in 3.5 cm dishes to approximately 80% confluency and then were treated with paracetamol as described above. Cells were washed with DPBS (1×2mL) and total RNA was extracted using the PureLink RNA mini kit (Invitrogen) following the manufacturer protocol. RNA was treated with DNase (Agilent) to degrade genomic DNA at 37 °C for 1 h and then with 5 mM EDTA at 75 °C for 10 min. Total RNA concentration was measured by nanodrop (DeNovix DS-11+). To synthetize cDNA, reverse transcription was performed using M-MLV (Moloney-Murine Leukemia Virus) reverse transcriptase (Invitrogen) added to 1.0 μg of total RNA for 50 min at 37 °C. Samples from reverse transcription were then diluted 1:10, aliquoted, and stored at -20 °C.

Quantitative Real-Time Polymerase Chain Reaction (RT-qPCR)
Samples consisting of ~10 ng template cDNA, forward and reverse primers (400-800 nM), 10 μL PowerUp TM SYBR Green master mix (ThermoFisher), and 7 μL nuclease free water were used to analyze gene expression levels by RT-qPCR on a LightCycler 480 (Roche). Nontemplate control (NTC) samples exchanging water for cDNA and control samples not subjected to reverse transcription (NRT) were also conducted. Experiments were carried out at 50 °C for 3 min and 95 °C for 5 min, followed by 75 cycles of 15 s at 95 °C and 30 s at 55 °C, followed by a melt curve analysis. After normalizing to housekeeping gene expression (ARP), the relative mRNA expression was calculated by the Livak method (2 −ΔΔCt ). Reported values represent the average of three biological replicates and were plotted as heatmaps using R(v 4.1.1). 18 Primers used are listed in Table S1.